fix(crypto): preserve bytes in latin1 digests

changelog.d/11029-crypto-latin1-digests.md

@@ -0,0 +1,6 @@
+### Fixed
+
+- `Hash.digest()` and `Hmac.digest()` now preserve every digest byte when
+  called with `"latin1"` or its `"binary"` alias. Bytes above `0x7f` no longer
+  become replacement characters, and the resulting string round-trips through
+  `Buffer.from(value, "latin1")` without corruption or length changes.

crates/perry-stdlib/src/crypto/hash_handles.rs

@@ -179,12 +179,16 @@ fn finalize_hmac_state(state: Option<HmacState>) -> Vec<u8> {
     }
 }
 
+fn latin1_string(bytes: &[u8]) -> String {
+    bytes.iter().map(|&byte| char::from(byte)).collect()
+}
+
 fn encoded_digest(bytes: &[u8], encoding: &str) -> String {
     match encoding {
         "hex" => hex::encode(bytes),
         "base64" => base64::engine::general_purpose::STANDARD.encode(bytes),
         "base64url" => base64::engine::general_purpose::URL_SAFE_NO_PAD.encode(bytes),
-        "binary" | "latin1" => String::from_utf8_lossy(bytes).into_owned(),
+        "binary" | "latin1" => latin1_string(bytes),
         _ => String::from_utf8_lossy(bytes).into_owned(),
     }
 }
@@ -498,7 +502,7 @@ pub unsafe fn dispatch_hash(handle: i64, method: &str, args: &[f64]) -> f64 {
                     "hex" => hex::encode(&digest),
                     "base64" => base64::engine::general_purpose::STANDARD.encode(&digest),
                     "base64url" => base64::engine::general_purpose::URL_SAFE_NO_PAD.encode(&digest),
-                    "binary" | "latin1" => String::from_utf8_lossy(&digest).into_owned(),
+                    "binary" | "latin1" => latin1_string(&digest),
                     _ => hex::encode(&digest),
                 };
                 let s = js_string_from_bytes(encoded.as_ptr(), encoded.len() as u32);
@@ -754,7 +758,7 @@ pub unsafe fn dispatch_hmac(handle: i64, method: &str, args: &[f64]) -> f64 {
                     "hex" => hex::encode(&digest),
                     "base64" => base64::engine::general_purpose::STANDARD.encode(&digest),
                     "base64url" => base64::engine::general_purpose::URL_SAFE_NO_PAD.encode(&digest),
-                    "binary" | "latin1" => String::from_utf8_lossy(&digest).into_owned(),
+                    "binary" | "latin1" => latin1_string(&digest),
                     _ => hex::encode(&digest),
                 };
                 let s = js_string_from_bytes(encoded.as_ptr(), encoded.len() as u32);

test-files/test_gap_10473_crypto_digest_latin1.ts

@@ -0,0 +1,15 @@
+import crypto from "node:crypto";
+
+function report(label: string, value: string, expectedHex: string) {
+  const codes = Array.from(value.slice(0, 6), (char) => char.charCodeAt(0)).join(",");
+  const roundTrips = Buffer.from(value, "latin1").toString("hex") === expectedHex;
+  console.log(label, value.length, codes, roundTrips);
+}
+
+const hashHex = crypto.createHash("sha256").update("abc").digest("hex");
+report("hash latin1", crypto.createHash("sha256").update("abc").digest("latin1"), hashHex);
+report("hash binary", crypto.createHash("sha256").update("abc").digest("binary"), hashHex);
+
+const hmacHex = crypto.createHmac("sha256", "k").update("abc").digest("hex");
+report("hmac latin1", crypto.createHmac("sha256", "k").update("abc").digest("latin1"), hmacHex);
+report("hmac binary", crypto.createHmac("sha256", "k").update("abc").digest("binary"), hmacHex);